Abstract
BACKGROUND: Overweight and obesity in mothers before pregnancy lead to overweight and obesity in their offspring, which is the main form of intergenerational transmission of overweight and obesity in early life. Many factors, especially non-genetic factors, may influence intergenerational transmission, but little prediction research has been conducted. Therefore, we analyzed the status of intergenerational transmission in maternal and infant overweight and obesity. Second, we explored the factors during the pregnancy that might affect the the intergenerational transmission; According to the two application scenarios of pregnancy screen and self-management, risk prediction models for pregnant women were carried out. METHODS: Based on a prospective birth cohort, a total of 908 mothers and offspring were followed up during early life. Follow-up visits were performed at the first trimester, second trimester, third trimester, delivery, 42 days after delivery, and 6 months and 12 months of age. The investigation methods included questionnaire survey, physical examination, biological sample collection and clinical data collection. In terms of risk prediction, univariate analysis was used to screen candidate predictors. Second, multivariable Cox proportional hazard regression models were used to determine the final selected predictors. Third, the corresponding histogram models were drawn, and then the 10-fold cross-validation methods were used for internal verification. RESULTS: Regarding intergenerational transmission of overweight and obesity between mothers and infants during pregnancy, the risk prediction model for pregnancy screen was constructed. The model established: h(t|X) = h(0)(t)exp.(- 0.95 × (Bachelor Degree or above) + 0.75 × (Fasting blood glucose in the second trimester) + 0.89 × (Blood pressure in the third trimester) + 0.80 × (Cholesterol in third trimester) + 0.55 × (Abdominal circumference in third trimester))., with good discrimination (AUC = 0.82) and calibration (Hosmer-Lemeshow(2) = 4.17). The risk prediction model for self-management was constructed. The model established: h(t|X) = h(0)(t)exp. (0.98 × (Sedentary >18METs) + 0.88 × (Sleep index≥8) + 0.81 × (Unhealthy eating patterns Q3/Q4) + 0.90 × (Unhealthy eating patterns Q4/Q4) + 0.85 × (Depression)), with good discrimination (AUC = 0.75) and calibration (Hosmer-Lemeshow(2) = 3.81). CONCLUSIONS: The risk predictions of intergenerational transmission of overweight and obesity between mothers and infants were performed for two populations and two application scenarios (pregnancy screening and home self-management). Further research needs to focus on infants and long-term risk prediction models.